Volume 12 • Number 4 • 2009 VA L U E I N H E A LT H
Health-Related Quality of Life (HRQL) and Continuous Antipsychotic Treatment: 3-year Results from the Schizophrenia Health Outcomes (SOHO) Study Jordi Alonso, MD, PhD,1,2 Tim Croudace, PhD,3 Jacqueline Brown, PhD,4 Isabelle Gasquet, MD,5 Martin R. J. Knapp, PhD,6,7 David Suárez, PhD,8 Diego Novick, MD4 1 Health Services Research Unit, Institut Municipal d’Investigació Mèdica (IMIM-Hospital del Mar), Barcelona, Spain; 2CIBER en Epidemiología y Salud Pública (CIBERESP), Barcelona, Spain; 3Department of Psychiatry, University of Cambridge, Addenbrooke’s Hospital, Cambridge, UK; 4 Eli Lilly and Company, Windlesham, Surrey, UK; 5Public Health Department-Paul Brousse Hospital (AP-HP),Villejuif, France; 6PSSRU, Social Policy Department, London School of Economics and Political Science, London, UK; 7Centre for the Economics of Mental Health, Institute of Psychiatry, London, UK; 8Sant Joan de Déu-SSM, Fundació Sant Joan de Déu, Sant Boi, Barcelona, Spain vhe_495
536..543
A B S T R AC T Objectives: We investigated the association between continuous antipsychotic use and health-related quality of life (HRQL) 3-year change in the European Schizophrenia Outpatients Health Outcomes (EU-SOHO) study. Methods: EU-SOHO is an observational study of outcomes associated with antipsychotic treatment for schizophrenia in an outpatient setting. HRQL was assessed at study entry and at 6, 12, 18, 24, 30, and 36 months using the EuroQol-5D (EQ-5D). UK population time trade-off (TTO) tariffs were applied to the self-rated EQ-5D health states to calculate HRQL ratings (0 = death, 1 = best). An epoch analysis approach was used as a conceptual framework to analyze the longitudinal data. Follow-up was divided into epochs or periods of continuous treatment. When a patient changed antipsychotic treatment, he or she was considered to have a new observation. Multilevel models were employed to evaluate the association of HRQL with medication and other clinical and
sociodemographic variables for each epoch. A total of 9340 patients were analyzed (42.1% women; mean age 40 years). Results: Mean EQ-5D scores increased over time; the largest improvement occurred in the first 6 months (mean increase of 0.19). Longer duration of illness and older age at first treatment were associated with worse baseline EQ-5D scores. Improvements in EQ-5D scores were greater for more socially active patients or those in paid employment. Few significant differences were found between antipsychotic medications. Olanzapine and clozapine were associated with higher HRQL increases. Conclusions: Continuous antipsychotic treatment is associated with important HRQL benefits at 3 years, most of which occurs during the first 6 months. Although some medications are associated with better HRQL outcomes, differences are small. Keywords: antipsychotics, health-related quality of life, outpatients, schizophrenia.
Introduction
to treatment but captures the impact of multiple relevant factors in a single global subjective judgment. HRQL measures have been applied in naturalistic settings and in clinical trials, but only the former setting characterizes the broader range of patients seen in everyday clinical practice. Because profiles of HRQL are reported by patients, not their clinicians, they are of interest to purchasers and providers of psychiatric services. A major problem in studying HRQL outcomes in response to antipsychotic treatment in observational studies is that patients with schizophrenia frequently change, or discontinue, their antipsychotic medication [12]. As a result, it is difficult to attribute changes in outcome to exposure to a particular drug [13] or relate differences in outcome solely to the pharmacological agents. Attributing improvements in clinical and HRQL outcomes to specific drug treatments requires analytic procedures which ensure that treatment intensity, duration, treatment history, and pretreatment prognostic factors are simultaneously taken into account. Hierarchical modeling of physician-related factors must also be included because patients receiving treatment from the same psychiatrist may experience similar outcomes due to factors such as treatment (optimal prescribing), insights developed into their illness, and the clinical relationship between patient and physician (therapeutic alliance). A number of methodological approaches can be applied to begin to address these complex issues, but how to do these effectively or optimally is still a matter of debate. Moreover, the results of observational studies have to be interpreted with caution because the observed
Treatment of schizophrenia with antipsychotic medication aims to improve not only the symptoms of illness but also the patient’s health-related quality of life (HRQL). New antipsychotic medications have been shown to improve clinical outcomes in the short term, both in clinical trials [1] and in observational studies [2]. Studies have analyzed the impact of antipsychotic treatment on HRQL, but the follow-up periods have been at most 1 year [3–9]. There is a lack of data on long-term effects of antipsychotic treatment on HRQL. Importantly, changes in HRQL associated with treatment could take longer to occur than clinical changes, or may be linked to the impact of clinical changes on social functioning, with HRQL improving as a consequence of both factors. Nevertheless, there is also an ongoing debate on whether stable patients with schizophrenia who have good symptom control would have a better HRQL if they were not maintained on continuous long-term antipsychotics [10]. HRQL is complex and influenced by many social, psychological, and clinical factors, including the patient’s age and sex, insight into illness, severity of current symptoms, and side effects of medications [11]. HRQL is not a simple measure of response Address correspondence to: Jordi Alonso, Health Services Research Unit, IMIM-Hospital del Mar, Parc de Recerca Biomèdica de Barcelona, Doctor Aiguader, 88. 08003 Barcelona, Spain. E-mail:
[email protected] 10.1111/j.1524-4733.2008.00495.x
536
© 2008, International Society for Pharmacoeconomics and Outcomes Research (ISPOR)
1098-3015/09/536
536–543
537
Quality of Life in Schizophrenia Outpatients outcome differences may be influenced by unmeasured confounding variables or selection to treatment effects [14,15]. Recent methodological developments may allow investigators to better control for biases present in observational studies [16]. The objectives of the present article were to describe the evolution of HRQL over a period of 3 years in outpatients with schizophrenia who initiated or changed to a new antipsychotic treatment at study entry, and to analyze the association between antipsychotic use and HRQL change within each period of continuous treatment, so that the effect of consecutive treatment periods (continuous treatment) could be explored. The European Schizophrenia Outpatients Health Outcomes (SOHO) study provided the context and all relevant data for the analyses.
Methods
Study Design and Selection Criteria The SOHO study is a 3-year longitudinal, observational study, conducted in 10 European countries (Denmark, France, Germany, Greece, Ireland, Italy, The Netherlands, Portugal, Spain, and the United Kingdom), of health outcomes associated with the treatment of schizophrenia. Ethical approval was obtained in all countries, either at the site, regional, or national level, depending on the country and local regulations. All patients gave at least oral informed consent, and written informed consent was obtained in Denmark, Italy, Ireland, Portugal, Spain, and the United Kingdom. Details of the rationale and design of the SOHO study have been provided elsewhere [17], as have the clinical results at 6, 12, and 36 months [18–20]. Recruitment. A total of 1096 participating psychiatrists enrolled 10,218 patients with schizophrenia who met the following entry criteria: initiating or changing antipsychotic medication for the treatment of schizophrenia, presenting within the normal course of care in the outpatient setting or in the hospital when admission was planned for the initiation of antipsychotic medication and discharge was planned within 2 weeks, aged at least 18 years, and not participating in an intervention study. Patients were included regardless of the reason for treatment change (e.g., lack of response, side effects, etc.) and regardless of whether an antipsychotic drug was being initiated as a replacement for a previous medication, whether it was an addition to existing treatment, or whether it was being initiated for the first time or after a period of no treatment. This was carried out to render the study broadly representative of schizophrenia outpatients. The recruitment period lasted from September 1, 2000 to December 31, 2001, and data collection finished in January 31, 2005. Cohorts. Through a stratified inclusion procedure that oversampled patients starting olanzapine, the study was designed to yield two patient cohorts of approximately equal size: 1) patients who initiated therapy with, or changed to, olanzapine; and 2) patients who initiated therapy with, or changed to, a nonolanzapine antipsychotic. Patients in the non-olanzapine cohort could be prescribed any medication chosen by the psychiatrist. The current analyses were restricted to those treatment groups with at least 100 patients initiating that antipsychotic at baseline and were as follows: olanzapine, risperidone, quetiapine, amisulpride, clozapine, oral typical antipsychotics, and depot typical antipsychotics. Patients who initiated two or more antipsychotics simultaneously were excluded from the analysis (N = 268) because the study was not intended to address the outcome of such joint prescribing/polypharmacy, and only a small fraction of the total sample received such prescriptions.
Enrolment. In most countries, enrollment was conducted in a systematic alternating order; e.g., such that the first patient was recruited into the olanzapine cohort, the second patient was recruited into the non–olanzapine cohort, and so on. Effort was made to avoid interference with normal clinical practice. Investigators were instructed to make treatment decisions independent of the study and then evaluate whether patients were eligible for inclusion based on the entry criteria and the alternating structure of enrollment. The enrollment period was purposely very long, and no minimum number of cases was required. All patient care during the study period was at the discretion of the participating psychiatrist because SOHO is a naturalistic observational study. In other words, they could prescribe and change medication at any time according to their usual clinical practice. No instructions about patient care were included in the study description. Data collection for the study occurred during visits that were within the normal course of health care. The routine outpatient visit at which patients were enrolled in the study served as the time for baseline data collection. Subsequent data collection was targeted for 3, 6, 12, 18, 24, 30, and 36 months postbaseline.
Outcome: Quality of Life HRQL was assessed using the EuroQol-5D (EQ-5D), a patient self-rated, generic, HRQL instrument that consists of two parts [21]: 1) a series of descriptive questions on general health covering the five health dimensions of mobility, self-care, usual activities, pain/discomfort, and anxiety/depression; and 2) a visual analog scale (VAS) that patients use to assess their current level of health on the day of scoring from 0 (worst imaginable health state) to 100 (best imaginable health state). The present study reports HRQL ratings using the former as it is more comprehensive. HRQL ratings for the self-rated EQ-5D health states were assigned using the readily available UK population tariffs (1 = best possible health; 0 = death) [22]. However, HRQL ratings using EQ-5D VAS were also analyzed. These gave similar results that are available upon request. Other measures. Other information collected included: sociodemographic variables (age, gender, country, employment, housing, and number of social activities) and clinical variables [baseline clinical global impression (CGI) score, whether the patients had ever received antipsychotic treatment for schizophrenia, age at first treatment contact, previous antipsychotic use, presence of extra-pyramidal symptoms, presence of tardive dyskinesia, CGI (positive, negative, depressive, cognitive) symptoms, body mass index at baseline, alcohol dependence, current substance abuse, previous hostile behavior, previous suicide attempts, previous compulsory admission or arrest, taking concomitant medication at baseline and antipsychotic medication].
Statistical Analysis Mean [and standard deviations (SD)] EQ-5D tariff scores are presented for each time point to describe the evolution of the HRQL. EQ-5D scores are also presented according to the baseline characteristics of the patients, recorded at study entry. Because the HRQL data are non-normally distributed (even after transformations are applied), nonparametric tests, such as the Wilcoxon signed-rank test, were used for simple group comparisons. Hierarchical linear regression models were used to model changes in HRQL because due to the large sample size and in the presence of covariates, this method improved assumptions about the normality of residuals and improved the robustness of the inferences about associations despite the distributional violations [23].
Alonso et al.
538 A total of 9340 patients with complete EQ-5D tariff information (91.4% of the total at study entry) were included in the analyses presented here. Epoch analysis of the longitudinal data. As is characteristic of schizophrenic patients [24], 42% of the patients in the SOHO study had either stopped or switched their baseline medication by 36 months. Difficulties arise with conventional methods of allocating and interpreting treatment effects on HRQL when patients frequently switch medication like this. With an intention-to-treat analysis approach, the recommended method of analyzing randomized controlled trials, the treatment effects are attributed to the medication given at study entry. When patients frequently change medication, however, this type of analysis could favor less-effective medications given at study entry if the patients subsequently switched to more effective medications. A method of analysis was thus needed to fully attribute the treatment effects on HRQL directly to the antipsychotic medication taken by the patients. An epoch analysis was developed and carried out; the full methodological details of which have been published elsewhere [25]. In essence, the 36-month follow-up period of the SOHO data set was considered as a series of six periods or epochs (0–6, 6–12, 12–18, 18–24, 24–30 or 30–36 months). The first 6 months of any new drug treatment initiated at any time during follow-up was considered the first epoch. If the same drug treatment was continued for another 6 months, then it would contribute data for the second epoch. Subsequent epochs were formed if the same drug treatment was continued for further 6-month intervals in a consecutive fashion. For instance, if a patient maintained the medication started at baseline for 24 months and then switched to a new medication that was maintained for another 12 months, he/she would have contributed to the analysis, with observations up to 24 months for the first medication (i.e., for the first four epochs each) and observations up to 12 months for the second medication (i.e., for the first two epochs each). This epoch approach was possible because patients entering the SOHO study were those who initiated or switched to a new antipsychotic at baseline. That is, the study is essentially studying a “window” in the patients’ treatment. If the patient’s medication was changed during the follow-up, then it was possible to consider their new treatment as a new observation equivalent to that at study entry. Thus, patients who changed medication during follow-up contributed to more than one treatment episode, with one treatment episode for each new medication started. This approach thus made the best use of the available data and took into account the longitudinal structure of the data, especially, antipsychotic-treatment switching, which is often due to the experience of a poor outcome. This is captured by including all the relevant health-status variables as covariates in the new epoch (i.e., they are “updated” from study baseline). Multivariate analyses. Six regression analyses were conducted, one for each follow-up period of continuous treatment. Each regression analysis assessed the factors associated with the change in HRQL at the end of the period. The covariates included in the model were HRQL, the clinical characteristics of the patient at the beginning of the period, the baseline clinical characteristics of the patient, and the medication taken during the period. A mixed modelling approach was used to take into account the multilevel and hierarchical nature of the data. Models included a random intercept at the investigator level (N = 985 investigators). Effects of country were addressed through a set of dummy variables, with Germany as the reference country. For the models with repeated observations per patient, a
random intercept at the patient level was also included to take into account the lack of independence of longitudinal repeated measures [26]. A similar approach has been described elsewhere and applied for cost-effectiveness and cost-utility analyses [27].
Results Of the 9340 patients assessed at study entry in the present analysis, the majority (58.1%) were evaluated at all subsequent study time points (Table 1). At study entry, patient mean age was 40 years (SD 13) and 42.1% were female. Mean duration of schizophrenia was 11.3 years (SD 11.0), and almost half of the sample (44.6%) reported having schizophrenia for at least 10 years. Most patients had their first contact with the health services for treatment of schizophrenia before the age of 35 years. Disease severity at study entry (as assessed using the CGI) was moderately severe (CGI = 4 in 36.7% patients) to severe (CGI = 5 in 35.2% patients). Table 1 summarizes the social and clinical characteristics of patients evaluated at all visits and other patients with some missing data. The groups were similar in terms of the characteristics measured. The baseline patient characteristics were also similar for the group of patients who completed the 36-month study (i.e., completers) and the group who dropped out at any time during follow-up (i.e., noncompleters) (data available in supplementary information, through ISPOR ViH supporting information). Table 2 summarizes the unadjusted EQ-5D tariff scores at study entry (baseline) and after 6 to 36 months for the total study sample and by patient characteristics. For the total study sample, baseline mean and median EQ-5D scores were 0.57 (SD 0.32) and 0.69, respectively. At baseline, there was a significant difference in the mean EQ-5D score according to CGI severity, ranging from 0.37 (SD 0.37) for patients with a CGI of 6 to 7, to 0.83 (SD 0.19) for those with a CGI score of 1 to 2 (P < 0.0001). Baseline mean EQ-5D scores increased significantly with an increasing number of social activities, from 0.48 (SD 0.35) among those with no social activities to 0.66 (SD 0.28) among those with five or more social activities during the past month (P < 0.0001). In addition, patients with paid employment had a significantly higher baseline mean EQ-5D score compared with those without paid employment; 0.64 (SD 0.30) versus 0.55 (SD 0.32), (P < 0.0001). Mean unadjusted EQ-5D scores increased from 0.57 (SD 0.32) at baseline to 0.80 (SD 0.25) after 36 months (Table 2). The increase in EQ-5D score was greatest after the first 6 months (mean increase of 0.19 for the total study sample). The increase in EQ-5D score was similar for most of the entry variables considered. An exception was found for the illness severity variable. Patients with the most severe disease (i.e., highest CGI score) at study entry had the greatest improvement in HRQL; mean EQ-5D score increased from 0.37 (SD 0.37) at study entry to 0.73 (SD 0.33) after 36 months (P < 0.0001). Nevetheless, those with the lowest (best) CGI score at entry achieved only little improvement: from 0.83 (SD 0.19) at study entry to 0.86 (SD 0.19) after 36 months (P = 0.006).
Factors Associated with HRQL Change: Results from Hierarchical, Conditional Models Multivariate analyses were performed to take into account relevant covariates at study entry as well as at the start of each new period of continued treatment. Table 3 summarizes the estimates for the change in EQ-5D tariff score between the specified levels of each variable for each of the six epochs. Duration of illness and age at first treatment contact were consistently associated with the change in EQ-5D score across all
539
Quality of Life in Schizophrenia Outpatients Table 1
Characteristics of the SOHO study sample at study entry, according to follow-up status
Total Gender Male Female Age at first contact (years) <20 20 to 25 25 to 30 30 to 35 35+ Duration of illness (years) Less than 1 1 to 5 5 to 10 10 to 20 20+ First episode Yes No CGI score 1 to 2, normal/minimally ill 3, mildly ill 4, moderately ill 5, markedly ill 6 to 7, severely/most severely ill Paid employment Yes No Independent residence Yes No Number of social activities during the past month 0 1 to 4 5+ Having a spouse or a partner Yes No Treatment initiated at entry Olanzapine Clozapine Risperidone Quetiapine Amisulpride Oral typical Depot typical
N (%) 9340 (100)*
1 to 4 visits N (%) 1645 (17.6)
5 to 7 visits N (%) 2272 (24.3)
All visits N (%) 5423 (58.1)
5364 (57.9) 3906 (42.1)
943 (57.8) 689 (42.2)
1311 (58.3) 937 (41.7)
3110 (57.7) 2280 (42.3)
1959 (22.2) 2254 (25.5) 1719 (19.4) 1065 (12.0) 1848 (20.9)
320 (20.9) 367 (24.0) 271 (17.7) 196 (12.8) 376 (24.6)
493 (23.0) 502 (23.4) 457 (21.3) 226 (10.5) 469 (21.8)
1146 (22.2) 1385 (26.8) 991 (19.2) 643 (12.4) 1003 (19.4)
1820 (21.0) 1581 (18.2) 1397 (16.1) 2065 (23.8) 1804 (20.8)
449 (30.1) 294 (19.7) 220 (14.8) 283 (19.0) 245 (16.4)
419 (20.1) 392 (18.8) 355 (17.1) 495 (23.8) 420 (20.2)
952 (18.7) 895 (17.6) 822 (16.1) 1287 (25.3) 1139 (22.4)
982 (10.5) 8337 (89.5)
234 (14.2) 1409 (85.8)
221 (9.8) 2044 (90.2)
527 (9.7) 4884 (90.3)
299 (3.2) 1222 (13.1) 3417 (36.7) 3282 (35.2) 1098 (11.8)
64 (3.9) 213 (13.0) 564 (34.4) 559 (34.1) 238 (14.5)
69 (3.0) 306 (13.5) 816 (36.0) 782 (34.5) 295 (13.0)
166 (3.1) 703 (13.0) 2037 (37.6) 1941 (35.9) 565 (10.4)
1832 (19.7) 7452 (80.3)
340 (20.8) 1296 (79.2)
426 (18.9) 1831 (81.1)
1066 (19.8) 4325 (80.2)
4543 (48.7) 4780 (51.3)
884 (53.8) 758 (46.2)
1113 (49.1) 1153 (50.9)
2546 (47.0) 2869 (53.0)
3001 (32.5) 4559 (49.4) 1667 (18.1)
552 (34.1) 748 (46.3) 317 (19.6)
807 (35.8) 1058 (46.9) 391 (17.3)
1642 (30,7) 2753 (51.4) 959 (17.9)
2717 (30.0) 6328 (70.0)
517 (32.5) 1075 (67.5)
670 (30.6) 1518 (69.4)
1530 (29.1) 3735 (70.9)
5068 (54.3) 306 (3.3) 1806 (19.3) 733 (7.9) 307 (3.3) 660 (7.1) 460 (4.9)
930 (56.5) 26 (1.6) 302 (18.4) 132 (8.0) 77 (4.7) 112 (6.8) 66 (4.0)
1210 (53.3) 90 (4.0) 434 (19.1) 196 (8.6) 68 (3.0) 165 (7.3) 109 (4.8)
2928 (54.0) 190 (3.5) 1070 (19.7) 405 (7.5) 162 (3.0) 383 (7.1) 285 (5.3)
*There are missing values for some categories, so totals do not add up to 9340 for some variables. CGI, clinical global impression; SOHO, Schizophrenia Outpatients Health Outcomes.
epoch durations when all other relevant covariates were held constant. Patients, having suffered from illness for more than 20 years, achieved 0.089 less change in EQ-5D score after 36-months continuous treatment compared with those having illness for less than 1 year. Similarly, patients who were older than 35 years at first contact for treatment had 0.056 less change in EQ-5D during 36-months continuous treatment compared with those who made the first contact for treatment when they were younger than 20 years old. In addition, having paid employment and having social activities were also associated with a greater change in EQ-5D (i.e., greater improvement) for most epoch durations. There were few differences in HRQL score changes between medication groups. Change in EQ-5D score was not significantly different between patients on olanzapine, amisulpride, and clozapine, but it was slightly lower for patients in all other treatment groups compared with the olanzapine group (Table 3). For example, patients taking olanzapine showed statistically significant greater improvements in EQ-5D score than patients taking oral typicals for the six epochs (-0.077, -0.078, -0.064, -0.072, -0.070, and -0.085, respectively).
Figure 1 presents the adjusted mean change (95% confidence interval) in EQ-5D tariff score after continuous treatment for 6 to 36 months according to duration of illness, baseline clinical severity, and baseline HRQL. Figure 1C shows that the 25% of schizophrenia patients with the worst HRQL at baseline [quartile (Q) 1, according to the EQ-5D score] experienced the largest improvement in their HRQL, while the 25% of patients with the best initial HRQL scores (Q4) did not experience any improvements throughout the whole epochs. In addition, the 25% of patients with second best initial HRQL scores (Q3) also experienced improvements only at the last two epochs (30 and 36 months of continuous treatment).
Discussion The present study has shown that antipsychotic treatment over a 36-month period is associated with substantial and sustained improvement in HRQL in outpatients with schizophrenia. The mean EQ-5D tariff score increased from 0.57 at study entry to 0.80 after 36 months, reaching a level very close to that of
Alonso et al.
540 Table 2
EQ-5D tariff score in the SOHO study by follow-up point, according to study entry characteristics
Total mean Total median (IQ range) Gender Male Female Age at first contact (years) <20 20 to 25 25 to 30 30 to 35 35+ Duration of illness (years) Less than 1 1 to 5 5 to 10 10 to 20 20+ First episode Yes No CGI score 1 to 2, normal/minimally ill 3, mildly ill 4, moderately ill 5, markedly ill 6 to 7, severely/most severely ill Paid employment Yes No Independent residence Yes No Number of social activities during the past month 0 1 to 4 5+ Spouse or partner Yes No
Baseline N = 9340
6 months N = 8374
12 months N = 7974
18 months N = 7454
24 months N = 6991
30 months N = 6566
36 months N = 6317
0.57 (0.32) 0.69 (0.50)
0.76 (0.26) 0.81 (0.31)
0.78 (0.25) 0.82 (0.29)
0.78 (0.26) 0.85 (0.29)
0.79 (0.25) 0.85 (0.29)
0.80 (0.25) 0.85 (0.29)
0.80 (0.25) 0.85 (0.27)
0.59 (0.31) 0.54 (0.33)
0.77 (0.25) 0.75 (0.27)
0.79 (0.24) 0.76 (0.27)
0.79 (0.25) 0.77 (0.26)
0.80 (0.24) 0.77 (0.27)
0.80 (0.24) 0.79 (0.26)
0.80 (0.25) 0.80 (0.25)
0.56 0.58 0.57 0.58 0.56
(0.33) (0.31) (0.32) (0.56) (0.33)
0.75 0.77 0.77 0.78 0.74
(0.26) (0.24) (0.25) (0.25) (0.28)
0.76 0.79 0.78 0.79 0.76
(0.26) (0.23) (0.24) (0.24) (0.27)
0.77 0.80 0.79 0.80 0.76
(0.26) (0.23) (0.25) (0.24) (0.27)
0.78 0.81 0.80 0.79 0.77
(0.26) (0.23) (0.25) (0.25) (0.28)
0.79 0.81 0.81 0.81 0.77
(0.25) (0.23) (0.25) (0.23) (0.27)
0.80 0.81 0.81 0.82 0.78
(0.25) (0.23) (0.25) (0.24) (0.26)
0.57 0.60 0.57 0.57 0.52
(0.31) (0.31) (0.31) (0.32) (0.34)
0.79 0.80 0.78 0.75 0.69
(0.23) (0.24) (0.24) (0.26) (0.29)
0.80 0.82 0.79 0.77 0.71
(0.23) (0.21) (0.24) (0.25) (0.29)
0.81 0.83 0.80 0.78 0.72
(0.24) (0.21) (0.24) (0.26) (0.29)
0.82 0.83 0.80 0.79 0.72
(0.23) (0.22) (0.24) (0.25) (0.29)
0.82 0.84 0.81 0.80 0.73
(0.23) (0.20) (0.24) (0.25) (0.28)
0.84 0.84 0.81 0.81 0.73
(0.22) (0.21) (0.24) (0.24) (0.29)
0.57 (0.32) 0.55 (0.31)
0.76 (0.26) 0.80 (0.22)
0.77 (0.25) 0.81 (0.24)
0.78 (0.26) 0.82 (0.23)
0.78 (0.26) 0.84 (0.21)
0.79 (0.25) 0.84 (0.21)
0.80 (0.25) 0.85 (0.20)
0.83 0.75 0.62 0.49 0.37
0.85 0.84 0.79 0.73 0.66
0.87 0.85 0.79 0.75 0.67
0.87 0.84 0.80 0.76 0.69
0.86 0.84 0.80 0.77 0.72
0.87 0.85 0.81 0.78 0.72
0.86 0.85 0.82 0.78 0.73
(0.19) (0.22) (0.28) (0.33) (0.37)
(0.21) (0.20) (0.22) (0.27) (0.32)
(0.18) (0.20) (0.23) (0.26) (0.32)
(0.18) (0.21) (0.23) (0.26) (0.33)
(0.19) (0.21) (0.23) (0.26) (0.31)
(0.17) (0.21) (0.24) (0.25) (0.32)
(0.19) (0.20) (0.22) (0.26) (0.33)
0.64 (0.30) 0.55 (0.32)
0.83 (0.21) 0.74 (0.26)
0.84 (0.21) 0.76 (0.26)
0.84 (0.22) 0.77 (0.26)
0.85 (0.21) 0.77 (0.26)
0.86 (0.20) 0.78 (0.25)
0.87 (0.20) 0.79 (0.26)
0.59 (0.31) 0.55 (0.33)
0.78 (0.25) 0.75 (0.26)
0.78 (0.25) 0.77 (0.26)
0.79 (0.25) 0.77 (0.26)
0.81 (0.25) 0.78 (0.26)
0.81 (0.24) 0.79 (0.26)
0.82 (0.24) 0.79 (0.26)
0.48 (0.35) 0.59 (0.30) 0.66 (0.28)
0.71 (0.29) 0.78 (0.24) 0.81 (0.22)
0.72 (0.29) 0.79 (0.23) 0.82 (0.21)
0.74 (0.29) 0.80 (0.24) 0.82 (0.22)
0.74 (0.29) 0.80 (0.23) 0.83 (0.22)
0.75 (0.29) 0.81 (0.23) 0.83 (0.22)
0.75 (0.29) 0.83 (0.22) 0.83 (0.23)
0.56 (0.32) 0.57 (0.32)
0.76 (0.26) 0.76 (0.25)
0.78 (0.26) 0.77 (0.25)
0.78 (0.26) 0.78 (0.25)
0.80 (0.25) 0.79 (0.25)
0.80 (0.25) 0.79 (0.25)
0.81 (0.25) 0.80 (0.25)
Data are unadjusted and values are expressed as mean (standard deviation) unless otherwise indicated. CGI, clinical global impression; EQ-5D, EuroQol-5D; IQ, interquartile; SOHO, Schizophrenia Outpatients Health Outcomes.
general European populations which range from 0.74 in Spain to 0.86 in the United Kingdom [28]. The improvements in EQ-5D tariff scores were higher during the first 6 months of treatment after study entry. However, these improvements were not homogenous for all the patients. For example, patients with the best initial HRQL scores did not experience improvements throughout continuous treatment. In addition, those with the second best initial HRQL scores only experienced improvements after 30 months of continuous treatment. This finding could partly be explained by regression to the mean of outcome scores for patients who were initially less ill. Shorter duration of illness and earlier age at first contact with health services were associated with better HRQL outcomes, as were being in paid employment or being more socially active at study entry. There were few differences in HRQL change among medication cohorts, but with this large sample size, we were able to detect statistically significant differences between olanzapine and the other medications (except clozapine and amisulpride), although the clinical significance of these differences may be marginal to moderate. However, the observational design limitations make it difficult to attribute outcome differences to individual drugs. Nevertheless, the sensitivity of our results to a range of such factors has been explored elsewhere and the conclusions altered very little [12]. Overall, the results are consistent with previous
studies of shorter duration, showing an important, continued improvement in HRQL associated with continuous therapy over 36 months. The SOHO study is a large, naturalistic, international, longterm follow-up study with a high retention rate. Moreover, the patients lost to follow-up were not substantially different from those retained in the study on the criteria available to us. The most important validity threat for SOHO is the consideration of medication changes and the lack of account taken of timevarying medication adherence (although most SOHO patients were highly adherent). Because patients change treatment during the follow-up, it would be unrealistic to assume that observed outcomes at follow-up are the consequence of initial treatment assignment. In most cases, a new treatment is indicated due to relapse, side effects, or failure to improve symptoms. It would be inappropriate to assign the outcomes to a previous treatment that had failed. The method used in the present analysis took into account medication changes. The 36-month results presented here are consistent with the shorter term data from randomized studies [9,29,30] which describe an important increase in HRQL after antipsychotic treatment. Moreover, long-term quality-of-life benefits of new antipsychotics have been reported recently; ziprasidone-treated patients had a better quality of life than haloperidol-treated
541
Quality of Life in Schizophrenia Outpatients Table 3
Hierarchical,‡ multivariate epoch analysis of the follow-up change in EQ-5D tariff score in the SOHO study Change in EQ-5D score for each epoch duration
Variable (reference category)
Category
6 months
12 months
18 months
24 months
30 months
36 months
Sex (female) Duration of illness (<1 year)
Male 1 to 5 5 to 10 10 to 20 20+ 20 to 25 25 to 30 30 to 35 35+ Obese Overweight Yes Yes Yes 1 to 4 5+ Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes NA NA NA NA 3 4 5 6 to 7 Yes Yes Yes Yes Yes Yes Yes 1 quartile 2 quartile 3 quartile Amisulpride Clozapine Depot typical Oral typical Quetiapine Risperidone
-0.005 0.007 -0.010 -0.017 -0.054** 0.000 -0.017 -0.010 -0.024* -0.005 -0.008 0.005 0.002 0.030** 0.020** 0.021* -0.001 0.009 -0.040* -0.004 -0.010 -0.014 0.008 -0.004 -0.026** -0.011 -0.003 -0.006 0.000 0.002 0.000 -0.003 0.001 -0.002 -0.010 -0.004 0.007 -0.021 0.015 0.017 -0.031 0.003 -0.024 0.530** 0.185** 0.081** -0.018 0.005 -0.064** -0.077** -0.043** -0.032**
-0.001 -0.005 -0.019 -0.025* -0.064** 0.002 -0.015 -0.007 -0.038** -0.005 0.009 0.008 0.003 0.023** 0.022** 0.024* 0.005 -0.007 -0.064** -0.010 -0.002 -0.024 0.004 0.010 -0.033** -0.006 -0.004 0.002 -0.002 -0.004 0.004 -0.003 -0.003 0.006 0.003 0.062 0.000 -0.021 0.009 0.005 0.008 -0.007 -0.004 0.560** 0.210** 0.085** -0.033 -0.003 -0.062** -0.078** -0.039** -0.040**
-0.002 -0.001 -0.012 -0.019 -0.060** 0.003 -0.012 -0.010 -0.043** -0.011 -0.004 0.004 0.003 0.011 0.014* 0.026* -0.008 -0.019* -0.048* -0.013 -0.013 -0.012 -0.005 0.001 -0.017 0.001 0.004 0.014 0.000 -0.002 0.002 -0.001 0.008 0.023 0.021 -0.004 0.001 0.000 0.014 -0.003 -0.020 -0.004 -0.020 0.570** 0.202** 0.085** -0.026 0.010 -0.055** -0.064** -0.031* -0.035**
0.003 -0.011 -0.015 -0.022 -0.079** -0.006 -0.030** -0.035* -0.050** -0.012 0.002 0.019* 0.013 0.014 0.013 0.019* 0.008 -0.011 -0.068** -0.005 0.000 -0.001 -0.002 0.024* -0.007 0.017 0.003 0.026 0.004 0.000 0.004 0.000 0.011 0.014 0.023 0.046 -0.006 -0.015 0.037* -0.018 -0.023 -0.014 -0.001 0.582** 0.225** 0.101** -0.035 0.011 -0.047** -0.072** -0.045** -0.026**
-0.010 -0.008 -0.030* -0.028* -0.074** -0.009 -0.018 -0.010 -0.037** -0.007 0.008 0.024** -0.002 0.026* 0.007 0.012 0.005 0.002 -0.039 -0.007 -0.001 0.006 -0.010 0.018 -0.025 0.017 0.005 0.025 -0.002 0.005 0.006 -0.003 -0.013 -0.006 0.029 0.132** -0.008 -0.012 0.028 0.016 -0.041 -0.006 -0.012 0.596** 0.250** 0.113** -0.036 0.009 -0.043* -0.070** -0.062** -0.030**
-0.009 -0.004 -0.028 -0.017 -0.089** -0.007 -0.023 -0.017 -0.056** 0.009 0.012 0.008 0.034** 0.025* 0.016 0.012 0.006 -0.030* -0.047* -0.002 -0.008 0.006 -0.001 0.025 -0.027 0.012 0.004 0.002 0.008 0.010* 0.007 -0.001 -0.026 -0.040* -0.019 0.111* -0.004 -0.018 0.028 0.008 -0.068* -0.009 0.018 0.623** 0.282** 0.105** -0.044 0.019 -0.090** -0.085** -0.031 -0.019
Age at first contact for treatment (<20 years)
BMI (under and normal)† Spouse (no) Independent housing (no) Paid employment (no) Number of social activities (zero activities) Previous oral typical use (no) Previous depot typical use (no) Previous clozapine use (no) Previous olanzapine use (no) Previous risperidone use (no) Previous other atypical use (no) First episode (no) Anticholinergics use (no) Antidepressant use (no) Anxiolytics/Hypnotic use (no) Mood stabilizer use (no) Any concomitant (no) Positive CGI Negative CGI Depressive CGI Cognitive CGI Overall CGI (1–2)
EPS (no) TD (no) Jail (no) Substance abuse (no) Alcohol dependence (no) Hostile (no) Suicide attempts (no) Baseline EQ-5D score (4 quartile base EQ-5D score) Treatment (olanzapine)
*P < 0.05; **P < 0.01 versus reference category. †BMI categories: under and normal (<25 kg/m2), overweight (25–30 kg/m2), and obese (>30 kg/m2). ‡ The participating investigator has been considered a random effect. BMI, body mass index; CGI, clinical global impression; EQ-5D, EuroQol-5D; EPS, extra-pyramidal symptoms; SOHO, Schizophrenia Outpatients Health Outcomes; TD, tardive dyskinesia.
patients during the 3-year double-blind extension phase of a 40-week randomized, double-blind study in patients with schizophrenia [31]. Our observation of greater improvement in HRQL with olanzapine and clozapine treatment compared with other antipsychotic medications is consistent with some previous studies [1,5,30,32–34] but not all [5,35]. One long-term naturalistic study also demonstrated an increase in HRQL for patients switching from conventional to novel or atypical antipsychotics [2]. Nevertheless, several recent studies found that atypical antipsychotics produced no greater improvement in quality of life compared to typical antipsychotics [36,37]. Furthermore, a large naturalistic study of Canadian patients with schizophrenia found that the improvement in quality of life over 2 years was not
influenced by the type of medication prescribed [38]. Although there are differences in the design and main outcome variables of these studies, we have ruled out the possibility that our own observations are biased due to switching antipsychotic treatment. Thus, maintained improvement in HRQL is attributable to the treatments studied. Some limitations of the study deserve discussion. First, due to the design of the study, half of the patients who were selected were starting olanzapine, rendering this a large treatment group. This, however, did not change the observational nature of the study, nor did it influence the interpretation of the results because the statistical analysis adjusted for the different sizes of the treatment cohorts. Importantly, any observed differences in baseline
Alonso et al.
542 (A)
(B)
(C)
䉳 Figure 1 Adjusted mean (95% confidence interval) change in EuroQol-5D (EQ-5D) tariff store after continuous treatment for 6–36 months in the Schizophrenia Outpatients Health Outcomes study according to patients’ duration of illness, severity of illness, and health-related quality of life at baseline. (A) Years of disease duration. (B) Overall baseline clinical global impression (CGI) score. (C) Baseline EQ-5D tariff score (quartiles).
characteristics between treatment groups were taken into account in the multivariate analyses. Also, we interpret magnitude of change rather than statistical significance. Second, selection bias might have been operating (even if involuntarily) to maximize the benefits of a particular treatment. Nevertheless, any such selection bias has been taken into account in the multivariate, multilevel analyses performed. In addition, it is important to emphasize that the outcome measure used (EQ-5D) was not assessed by the clinician but was self-rated by the patient. Finally, the main HRQL instrument used in this study, EQ-5D, has been used in many other studies, but in few clinical trials assessing the effectiveness of antipsychotic treatment. This makes the results difficult to compare with previous studies. Nevertheless, the EQ-5D is widely used in Europe [21,28], the United States [39], and other countries [21]. Based on these data, it is clear that the patients with schizophrenia in the SOHO study had an impaired HRQL at study entry and that continuous treatment was associated with an important, maintained improvement in HRQL that approached that of the general population average. In summary, we present data on longer term HRQL based on a naturalistic follow-up study that has taken into account the medication changes occurring during the 3-year observation period. Our results show important and continued benefit associated with antipsychotic treatment in outpatients with schizophrenia. The results also illustrate the need to take into account all medication changes and other variables that associated with a change in HRQL. Source of financial support: Eli Lilly and Co. makes a donation to research funds in the Institut Municipal d’Investigació Mèdica of Barcelona that Jordi Alonso gratefully acknowledges, and in the Department of Psychiatry of the University of Cambridge that Tim Croudace gratefully acknowledges. David Suárez and Martin Knapp have received economic compensation for their participation in the SOHO Advisory Board. Jacqueline Brown and Diego Novick are Lilly employees. David Suárez was supported by a research grant from the Instituto de Salud Carlos III, Ministerio de Sanidad y Consumo, Spain (FIS CA05/0177).
Supporting Information Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.
References 1 Davis JM, Chen N, Glick ID. A meta-analysis of the efficacy of second generation antipsychotics. Arch Gen Psychiatry 2003;60: 553–64. 2 Voruganti L, Cortese L, Owyeumi L, et al. Switching from conventional to novel antipsychotic drugs: results of a prospective naturalistic study. Schizophr Res 2002;57:201–8. 3 Revicki DA, Genduso LA, Hamilton SH, et al. Olanzapine versus haloperidol in the treatment of schizophrenia and other psychotic disorders: quality of life and clinical outcomes of a randomized clinical trial. Qual Life Res 1999;8:417–26.
Quality of Life in Schizophrenia Outpatients 4 Tunis SL, Croghan TW, Heliman DK, et al. Reliability, validity and application of the medical outcome study SF-36 short form health survey (SF-36) in schizophrenic patients treated with olanzapine versus haloperidol. Med Care 1999;37:678–91. 5 Rosenheck R, Perlich D, Bingham S, et al. Effectiveness and cost of olanzapine and haloperidol in the treatment of schizophrenia: a randomized controlled trial. JAMA 2003;290:2693–702. 6 Ristner M, Gible A, Perelroyzen G, et al. Quality of life outcomes of risperidone, olanzapine, and typical antipsychotics among schizophrenia patients treated in routine clinical practice: a naturalistic comparative study. J Clin Psychopharmacol 2004;24:582– 91. 7 Kilian R, Dietrich S, Toumi M, et al. Quality of life in persons with schizophrenia in outpatient treatment with first- or secondgeneration antipsychotics. Acta Psychiatr Scand 2004;110:108– 18. 8 Naber D, Riedel M, Klimke A, et al. Randomized double blind comparison of olanzapine vs. clozapine on subjective well-being and clinical outcome in patients with schizophrenia. Acta Psychiatr Scand 2005;111:106–15. 9 Strakowski SM, Johnson JL, DelBello MP, et al. Quality of life during treatment with haloperidol or olanzapine in the year following a first psychotic episode. Schizophr Res 2005;78:161–9. 10 Beasley CM, Sutton VK, Taylor CC, et al. Is quality of life among minimally symptomatic patients with schizophrenia better following withdrawal or continuation of antipsychotic treatment? J Clin Psychopharmacol 2006;26:40–4. 11 Hofer A, Kemmler G, Eder U, et al. Quality of life in schizophrenia: the impact of psychopathology, attitude toward medication and side effects. J Clin Psychiatry 2004;65:932–9. 12 Haro JM, Suárez D, Novick D, et al. Three-year antipsychotic effectiveness in the outpatient care of schizophrenia: observational versus randomized studies results. Eur Neuropsychopharmacol 2007;17:235–44. 13 Covell NH, Frisman LK, Essock SM. Rates of refusal to participate in research studies among men and women. Psychiatr Serv 2003;54:1541–4. 14 D’Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:2265–81. 15 Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41–55. 16 Haro JM, Kontodimas S, Negrin MA, et al. Methodological aspects in the assessment of treatment effects in observational health outcome studies. Appl Health Econ Health Policy 2006; 5:11–25. 17 Haro JM, Edgell ET, Jones PB, et al. The European Schizophrenia Outpatient Health Outcomes (SOHO) study: rationale, methods and recruitment. Acta Psychiatr Scand 2003;107:222–32. 18 Haro JM, Edgell ET, Novick D, et al. Effectiveness of antipsychotic treatment for schizophrenia: 6-month results of the PanEuropean Schizophrenia Outpatient Health Outcomes (SOHO) study. Acta Psychiatr Scand 2005;111:220–31. 19 Haro JM, Novick D, Belger M, Jones PB. Antipsychotic type and correlates of antipsychotic treatment discontinuation in the outpatient treatment of schizophrenia. Eur Psychiatry 2006;21:41–7. 20 Haro JM, Novick D, Suárez D, et al. Remission and relapse in the outpatient care of schizophrenia: three-year results from the Schizophrenia Outpatient Health Outcomes study. J Clin Psychopharmacol 2006;26:571–8. 21 Brooks R, Rabin R, de Charro F. The Measurement and Valuation of Health Status Using EQ-5D: A European Perspective. Dordrecht, The Netherlands: Kluwer Academic Publishers, 2003. 22 Kind P, Hardman G, Macran S. UK Population Norms for EQ-5D [Working Paper #172]. York, UK: York Centre for Health Economics, 1999.
543 23 Lumley T, Diehr P, Emerson S, Chen L. The importance of the normality assumption in large public health data sets. Annu Rev Public Health 2002;23:151–69. 24 Covell NH, Jackson CT, Evans AC, Essock SM. Antipsychotic prescribing practices in Connecticut’s public mental health system: rates of changing medications and prescribing styles. Schizophr Bull 2002;28:17–29. 25 Windmeijer F, Kontodimas S, Knapp M, et al. Methodological approach for assessing the cost-effectiveness of treatments using longitudinal observational data: the SOHO study. Int J Technol Assess Health Care 2006;22:460–8. 26 Laird NM, Ware JH. Random-effects models for longitudinal data. Biometrics 1982;38:963–74. 27 Knapp M, Windmeijer F, Brown J, et al. Cost-utility analysis of treatment with olanzapine compared with other antipsychotic treatments in patients with schizophrenia in the pan-European SOHO study. Pharmacoeconomics 2008;26:341–58. 28 Szende A, Williams A. Measuring Self Reported Population Health: An International Perspective Based on the EQ-5D. Budapest, Hungary: SpringMed Publication Ltd, 2004. 29 Silva de Lima M, de Jesús Mari J, Breier A, et al. Quality of life in schizophrenia: a multicenter, randomized, naturalistic, controlled trial comparing olanzapine to first generation antipsychotics. J Clin Psychiatry 2005;66:831–8. 30 Lieberman JA, Scott Stroup T, McEvoy JP, et al. Effectiveness of antipsychotic drugs in patients with chronic schizophrenia. N Engl J Med 2005;353:1209–23. 31 Loebel AD, Warrington L, Siu C, Lieberman JA. Remission in schizophrenia: a comparison of 2 dose regimens of ziprasidone vs. haloperidol treatment in a 40-week core and 3-year double-blind extension study. Abstract NR395. Program and Abstracts of American Psychiatric Assoc Annual Meeting, May 20–25 2006, Toronto, Ontario, Canada, 2006. 32 Ritsner MS, Gibel A. The effectiveness and predictors of response to antipsychotic agents to treat impaired quality of life in schizophrenia: a 12-month naturalistic follow-up study with implications for confounding factors, antidepressants, anxiolytics and mood stabilisers. Prog Neuropsychopharmacol Biol Psychiatry 2006;30:1442–52. 33 Lewis SW, Barnes TR, Davies L, et al. Randomised controlled trial of effect of prescription of clozapine versus other secondgeneration antipsychotic drugs in resistant schizophrenia. Schizophr Bull 2006;32:715–23. 34 Mayoral F, Montejo AL, Bousono M, et al. [Quality of life and social functioning in schizophrenic patients treated with olanzapine: 1 year follow-up naturalistic study]. Actas Esp Psiquiatr 2006;34:7–15. [Spanish]. 35 Geddes J, Freemantale N, Harrison P, Bebbington P. Atypical antipsychotics in the treatment of schizophrenia: systematic overview and meta-regression analysis. BMJ 2000;321:1371–6. 36 Kilian R, Angermyer MC. The effects of antipsychotic treatment on quality of life of schizophrenic patients under naturalistic conditions: an application of random effect regression models and propensity scores in an observational prospective trial. Qual Life Res 2005;14:1275–89. 37 Jones PB, Barnes TR, Davies L, et al. Randomised controlled trial of the effect on quality of life of second- vs. first-generation antipsychotic drugs in schizophrenia. Cost Utility of the Latest Antipsychotic drugs in Schizophrenia (CUtLASS 1). Arch Gen Psychiatry 2006;63:1079–87. 38 Malla A, Williams R, Kopala L, et al. Outcome on quality of life in a Canadian national sample of patients with schizophrenia and related psychotic disorders. Acta Psychiatr Scand 2006; 430(Suppl.):S22–8. 39 Johnson JA, Coons SJ. Comparison of the EQ-5D and SF-12 in an adult SF12 sample. Qual Life Res 1998;7:155–66.